Deep Learning with Differential Privacy
Open accessThe DOI matches its Crossref record; the references it cites are not checked yet.
Abstract
Machine learning techniques based on neural networks are achieving remarkable results in a wide variety of domains. Often, the training of models requires large, representative datasets, which may be crowdsourced and contain sensitive information. The models should not expose private information in these datasets. Addressing this goal, we develop new algorithmic techniques for learning and a refined analysis of privacy costs within the framework of differential privacy. Our implementation and experiments demonstrate that we can train deep neural networks with non-convex objectives, under a modest privacy budget, and at a manageable cost in software complexity, training efficiency, and model quality.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source doesn’t support the claim as written.
In the citing paper
“Reference [24] added Gaussian noise on the clipped gradient but did not explain how to select the clipping threshold.”
From A Blockchain‐Based Federated Learning Method for Smart Healthcare (2021) · cites this paper as [24]
Checked by a reasoning model — may be wrong.
Experimental — accuracy not yet measuredThe cited source doesn’t support the claim as written.
In the citing paper
“To this end, DP injects random noise to the data before it is processed.”
From Connecting the dots in trustworthy Artificial Intelligence: From AI principles, ethics, and key requirements to responsible AI systems and regulation (2023) · cites this paper as [49]
Checked by a reasoning model — may be wrong.
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“However, it has been shown that the models themselves can, under certain conditions, memorise information40–43.”
From The future of digital health with federated learning (2020) · cites this paper as [44]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“We note that memorization can be mitigated using components in FL such as DP [25-28] (as discussed later).”
From A Pragmatic Framework for Federated Learning Risk and Governance in Academic Medical Centers (2026) · cites this paper as [25]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“While current efforts try to increase the privacy of federated learning through the use of mechanisms such as differential privacy [190] and secure multiparty computation [11], these approaches often provide privacy at the cost of lesser system efficiency or reduced model accuracy.”
From Applications of Federated Learning; Taxonomy, Challenges, and Research Trends (2022) · cites this paper as [190]
16 more claims about this paper
Create a free account to see every claim other papers make about this one.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
The citation lists are queued
They are fetched later and appear here then.
References 64 · Cited by 6,578
Research works printed but not linked: 5
- M. Bun and T. Steinke . Concentrated di erential privacy: Simpli cations, extensions, and lower bounds. CoRR, abs/1605.02065 , 2016 . M. Bun and T. Steinke. Concentrated di erential privacy: Simpli cations, extensions, and lower bounds. CoRR, abs/1605.02065, 2016. · arXiv
- 10.5555/1953048.2021036 · doi:10.5555/1953048.2021036
- 10.5555/1873601.1873691 · doi:10.5555/1873601.1873691
- P. Kairouz , S. Oh , and P. Viswanath . The composition theorem for di erential privacy . In ICML , pages 1376 -- 1385 . ACM, 2015 . P. Kairouz, S. Oh, and P. Viswanath. The composition theorem for di erential privacy. In ICML, pages 1376--1385. ACM, 2015.
- 10.5555/646765.704129 · doi:10.5555/646765.704129
Metadata
- Authors
- Abadi, M., Chu, A., Goodfellow, I., McMahan, H. B., Mironov, I., Talwar, K., & Zhang, L.
- Affiliations
- Google, Mountain View, CA, USA
- Google (United States) (US)
- OpenAI (United States) (US)
- Venue
- Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security
- Publisher
- ACM
- Published
- 2016-10-24
- Pages
- 308-318
- ISSN
- 1543-7221 (ISSN-L)
- OpenAlex
- W2473418344
- Cited by
- 6,578
- References
- 64
- Citations by year
- 2026: 958 · 2025: 1088 · 2024: 994 · 2023: 880 · 2022: 654 · 2021: 772 · 2020: 653 · 2019: 347 · 2018: 150 · 2017: 70 · 2016: 9
- FWCI
- 101.52
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- ACM SIGSAC Conference on Computer and Communications Security (CCS) (publishedVersion, cc-by-nc-nd)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
- Physical Sciences › Computer Science › Artificial Intelligence › Stochastic Gradient Optimization Techniques
- Physical Sciences › Computer Science › Computer Networks and Communications › Cryptography and Data Security
- Keywords
- differential privacy, deep neural network, privacy-preserving machine learning, non-convex optimization, privacy budget, sensitive data protection, training efficiency
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Deep Learning with Differential Privacy. https://app.cochat.ai/papers/10.1145/2976749.2978318
- The paper
- Abadi, M., Chu, A., Goodfellow, I., McMahan, H. B., Mironov, I., Talwar, K., & Zhang, L. (2016). Deep Learning with Differential Privacy. Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security. https://doi.org/10.1145/2976749.2978318
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.